Many of the targets for pharmaceutical drug discovery are ligands for receptor proteins, many of which have recently been cloned and pharmacologically characterized. Now that a large number of receptors have been cloned, a major goal of the pharmaceutical industry is to identify ligands for these receptors by screening vast libraries of substances. Unfortunately, with available methods and technology, a major limitation in the drug discovery process is the time and expense required to screen these libraries against so many targets.
The first step in the characterization of ligand interaction with a cloned receptor is to express the receptor in a ligand sensitive form. While a few receptors can be expressed in easily manipulated model systems such as yeast and E. coli, the interactions of ligands with most receptors are influenced by postranslational modifications that are only present in mammalian cells, and many of these receptors require mammalian proteins to accurately transduce their biological effects. Thus for wide applicability, an assay system must be based on expression of cloned receptors in mammalian cells.
The ability of ligands to interact with receptors can be evaluated by competition with a labeled ligand (e.g. radionucleotide) for a binding site on the receptor. Such assays are popular because they involve relatively few steps. Also, since binding often does not require interaction with other cellular proteins, these assays are less sensitive to factors such as levels of expression of the receptor and the cellular environment. Recently, technology such as the Proximity Assay (Amersham Co.) has further simplified these assays making automation and mass screening possible. Binding assays have many limitations: (i) For many technical reasons, binding assays are almost always performed in nonphysiological buffers. These buffers often markedly influence receptor pharmacology. (ii) Agonists and antagonists are not reliably discriminated in binding assays. (iii) Only binding sites for which labeled ligands are available can be studied. (iv) Since only modest levels of receptor (binding site) expression have been achieved in mammalian cells, propagation of receptors is a major expense in these assays. (v) The vast majority of labeled ligands are radioisotopes, the purchase, handling and disposal of which are major expenses.
To reliably discriminate between agonist and antagonist ligands, a response of the receptor must be measured. Responses to agonist activation of receptors are commonly measured as altered activity of various endogenous cellular proteins. Examples include measurement of second messengers such as cAMP (adenylyl cyclase), phosphoinositol metabolism (phospholipase c), tyrosine phosphorylation, and ion channels. All of these assays require the use of cells and/or cellular preparations that have a high degree of biological integrity, and these assays include many complex and expensive steps (Schlessinger and Ullrich, Neuron 9, 383 (1992); chapters in Molecular Biology of G-protein-coupled receptors, M. Brann, ed., Birkhauser (1992)).
A strategy that has been used to avoid the time and expense of measurement of endogenous proteins is to express conveniently assayed marker proteins that can be controlled by activation of the receptor. For example, receptors that control levels of transcription factors can be assayed using markers whose expression is under the transcriptional control of these factors. While this approach has led to convenient assays of receptors that are known to function as controllers of transcription (e.g. steroid/thyroid hormone receptors, Evans (WO 91/07488); Spanjaard et al. Mol. Endocrinology  7:12-16 (1993)), these assays have proven to have limited utility when applied to cell surface receptors, presumably because of the more modest transcriptional control that these receptors exert. Other than the assays that are based on transcriptional control, no approach has been described to assay receptors via recombinant markers that can be conveniently measured.
Another approach is to express the receptors in specialized cells that have endogenous response mechanisms that allow convenient assay of ligand activation of the receptor. Two examples include the RBL cells and melanophores. In RBL cells, muscarinic receptors that stimulate phospholipase c enhance the release of the enzyme hexosaminidase (Jones et al., FEBS Lett. 289, 47 (1991)), a conveniently measured response. In melanophores (cultured pigment cells) cloned receptors that change cAMP levels alter cellular color, a response that is similarly easily measured (Potenza et al., Anal. Biochem. 206, 315 (1992)). The limitations of these assays are that only certain functional types of receptors can be measured. Also, while the assays are relatively convenient, there are limitations inherent in the endogenous responses and cells that are used.
When exposed to ligands, a wide diversity of receptors are able to alter the pH of the media that is used for cell culture. These pH changes are small in magnitude and require expensive instrumentation for measurement (Cytosensor, Molecular Dynamics Co.). This device is not compatible with other instruments that are used in mass screening (e.g. use of a 96 well plate format) and because samples must be incubated within the instrument for several minutes, there is limited sample throughput.
A theoretical limitation inherent in all of the above assays is the inability to assay a given ligand against more than a few receptors at the same time. For example, radioligand binding assays can only be multiplexed to the extent that different and distinguishable radioisotopes are available (e.g. 3H versus 125I). Because of their limited dynamic range, incompatible assay conditions, and the fact that many receptors cannot be distinguished from one another based on their functional responses, second messenger responses, and most other biochemical effects of receptors, are not at all amenable to multiplexed assay. Similarly, the RBL assay, melanophore assay, and Cytosenor pH assays, are only applicable to assay of a single receptor at a time.
Another cellular response that is shared by many receptors is the ability to alter cellular growth. NIH 3T3 cells are a fibroblast cell line that has been extensively used to evaluate the activity of large diversity of gene products that control cell growth, and a number of receptors are able to control the activity of these cells when stimulated by individual ligands. Examples include nerve growth factor (NGF) which stimulates growth only when these cells have been transfected with trk A receptors (NGF receptor) (Cordon-Cardo et al., Cell 66:173-183 (1992); Chao, Neuron 9:583-593 (1992)), carbachol (a muscarinic agonist) stimulates cells transfected with certain muscarinic receptors (Gutkind et al., Proc. Natl. Acad. Sci. USA 88, 4703 (1991); Stephens et al., Oncogene 8, 19-26 (1993)), and nonpinephrine stimulates cells transfected with certain a adrenergic receptors (Allen et al., Proc. Natl. Acad. Sci. USA 88, 11354 (1991)). After long-term stimulation with agonist ligands, the cells change a number of characteristics including cellular growth, loss of contact inhibition, and formation of macroscopic colonies called foci. The ability to induce foci in NIH 3T3 cells is a common characteristic of cancer-associated genes (oncogenes).
The ability of receptors and other gene products to stimulate growth and induce foci in NIH 3T3 cells correlates with the stimulation of individual second messenger systems. Trk A receptors stimulate tyrosine phosphorylation (tyrosine kinase receptor), and many other genes that stimulate tyrosine phosphorylation stimulate growth and focus production in NIH 3T3 cells (Schlessinger and Ullrich, Neuron 9, 383 (1992)). Certain muscarinic (Gutkind et al., Proc. Natl. Acad. Sci. USA 88, 4703 (1991)), adrenergic (Allen et al., Proc. Natl. Acad. Sci. USA 88, 11354 (1991)) and serotonergic (Julius et al., Science 244, 1057 (1989)) receptors that stimulate phospholipase c, also stimulate growth and focus formation in NIH 3T3 cells. In the case of the muscarinic receptors, the ability to stimulate foci and phospholipase c have exactly the same dose/response characteristics, suggesting that these responses may be used as assays for ligand interactions. Unfortunately, these assays offer few advantages to the approaches described above. Focus assays involve a response that requires at least two weeks of cell culture, and are confounded by qualitative changes in patterns of growth. Direct measurement of cellular growth has also been used to measure effects of ligands. The most commonly used assay is 3H-thymidine incorporation (Stephens et al., Oncogene 8, 1993, pp. 19-26). These assays are neither convenient nor inexpensive to perform.
Schizophrenia is a devastating neuropsychiatric disorder that affects approximately 1% of the human population. It is characterized by a constellation of symptoms: “positive” symptoms such as hallucinations and delusions; and “negative” symptoms such as social and emotional withdrawal, apathy, and poverty of speech. The disorder usually develops early in life and is characterized by a chronic, often relapsing remitting course. Although the pathophysiology of this clinically heterogeneous disorder is unknown, genetic factors play a significant role. It has been estimated that the total financial cost for the diagnosis, treatment, and lost societal productivity of individuals affected by this disease exceeds 2% of the gross national product (GNP) of the United States. To date, there exist no definitive diagnostic tests for this disorder. Current treatment options available to psychiatrists primarily involve pharmacotherapy with a class of drugs known as antipsychotics. Antipsychotics are effective in ameliorating positive symptomotology, yet they frequently do not improve negative symptoms, and significant, treatment-limiting side effects are common.
Drugs that possess antipsychotic properties have been in clinical use since the early 1950's. The first compound shown to possess this property was chlorpromazine, and many of the subsequent compounds were derived from this phenothiazine antipsychotic. Currently, nine major classes of antipsychotics have been developed and are widely prescribed to treat psychotic symptoms irrespective of their etiology. Clinical use of these compounds are limited, however, by their side effect profiles. Nearly all of the “typical” or older generation compounds have significant adverse effects on human motor function. These “extrapyramidal” side effects, so termed due to their effects on modulatory human motor systems, can be both acute and chronic in nature. Acute effects include dystonic reactions, and a potentially life threatening but rare symptom constellation, neuroleptic malignant syndrome. Chronic side effects include akathisias, tremors, and tardive dyskinesia, a movement disorder characterized by involuntary writhing movements of the tongue and oral musculature seen with long-term administration of these agents. Due in large part to these disabling side effects, drug development in this class of compounds has been focused on newer “atypical” agents free of these adverse effects.
Various hypotheses have been proposed concerning the pathophysiology of schizophrenia, including genetic, environmental, and developmentally based theories. Current neuropharmacological theories are based, in large part, on the observation that antipsychotic drugs can improve the symptoms of schizophrenia, coupled with our current knowledge as to the mechanism of action of this class of drugs. Antipsychotic drugs have been shown, by both in vitro and in vivo methods, to interact with a large number of central monoaminergic neurotransmitter receptors, including dopaminergic, serotonergic, adrenergic, muscarinic, and histaminergic receptors. It is likely that the therapeutic and adverse effects of these drugs are mediated by distinct receptor subtypes.
The prevailing theory as to the mechanism of action of antipsychotic drugs involves antagonism of dopamine D2 receptors. This is based on the observation that these drugs have high affinity for this receptor in vitro, and that a correlation exists between their potency to block D2 receptors and their clinical efficacy. Unfortunately, it is likely that antagonism of dopamine D2 receptors also mediates the disabling extrapyramidal side effects. Interestingly, some antipsychotic drugs have been shown not to possess high affinity for D2 receptors, and therefore an alternate mechanism must be responsible for their clinical effects. The only other consistent receptor interaction that these drugs as a class display is antagonism of 5-HT2A receptors, suggesting that antagonism of these receptors is an alternate molecular mechanism that confers antipsychotic efficacy.
The observation that many of these drugs are antagonists of 5-HT2A receptors has led investigators to postulate that schizophrenia might be caused by heightened or exagerrated signal transduction through serotonergic systems. This theory is bolstered by a number of basic scientific and clinical observations regarding serotonergic systems and the 5-HT2A receptor in particular. Firstly, in addition to the known antipsychotics in widespread clinical usage, research compounds (e.g. ritanserin) that-selectively block 5-HT2A receptors (with respect to D2 receptors) have also been shown to possess antipsychotic activity. Secondly, the 5-HT2A receptor mRNA and protein have been shown to be expressed in neural systems that mediate higher cognitive and affective functions, including the cerebral cortex, hippocampus, and amygdala. Thirdly, some of the positive symptoms that characterize the disease can be mimicked in normal individuals by the ingestion of the hallucinogenic indolamine lysergic acid diethylamide (LSD). It is known that LSD and similar hallucinogens exert their psychogenic effects, in part, through the activation of 5-HT2A receptors. G-protein coupled neurotransmitter receptors (GPCR's), including the 5-HT2A receptor, function as transducers of intercellular communication. Traditionally, these receptors have been assumed to exist in a quiescent state unless activated by the binding of an agonist (a drug that activates a receptor). When activated, receptors interact with G-proteins, resulting in the generation, or inhibition of, second messenger molecules such as cyclic AMP, inositol phosphates, and diacylglycerol. These second messengers then modulate the function of a variety of intracellular enzymes, including kinases and ion channels, which ultimately determine neuronal excitability and neurotransmitter release.
Over the last few years some fundamental observations have been made relating to ways in which these receptor molecules function. One of the most important of these has been the identification and characterization of constitutively active receptors. It is now appreciated that many, if not most, of the GPCR monoamine receptors can exist in a partially activated state in the absence of their agonists. This increased basal activity can be inhibited by a class of drugs aptly named inverse agonists, in that they function as the inverse of agonists. Inverse agonists differ mechanistically from classic (or neutral) antagonists. Antagonists compete against agonists and inverse agonists for access to the receptor, but do not possess the intrinsic ability to inhibit elevated basal or constitutive receptor responses.
Multiple lines of experimental evidence support the hypothesis that constitutively active neurotransmitter receptors may exist in the central nervous system and be causative for human neuropsychiatric disease. Constitutive activity has been observed with neurotransmitter receptors mutated in vitro. For instance, S. Cottechia et al. (Proc. Natl. Acad. Sci. USA 87, 1990, pp. 2896-2900) made constitutively active chimeric α-1 adrenergic receptor by replacing the third intracytoplasmic loop of the receptor with that of the β-2 adrenergic receptor. Also, P. Samama et al. (J. Biol. Chem. 268, 1993, pp. 4625-4636) generated a constitutively active P2 receptor by replacing four amino acid residues in the C-terminal region of the third intracytoplasmic loop with analogous residues from the α-1B receptor. Point mutations have been introduced into the muscarinic m5 receptor by random saturation mutagenesis (E. S. Burstein et al., Biochem. Pharmacol. 51, 1996, pp. 539-544; T. A. Spalding et al., J. Pharm. Exp. Ther. 275, 1995, pp. 1274-1279), resulting in more than 40 mutants that exhibit varying degrees of constitutive activity. The relative ease with which these receptors may be mutated to a constitutively active form suggests that constitutively active receptors may occur spontaneously in nature with a high frequency.
A strong argument for the potential contribution of constitutively active receptors to human neuropsychiatric disease would be the finding that similar mutations are causative in other human diseases. Mutations in the G-protein coupled receptor gene family are common and are increasingly recognized to cause a number of human diseases. Most of these mutations are single nucleotide or point mutations that alter the structure and function of the receptor molecules. For instance, point mutations in the receptors rhodopsin and vasopressin (J. Nathans, Cell 78, 1994, pp. 357-360; W. Rosenthal et al., Nature 359, 1992, pp. 233-235) cause reading frame shifts, prematurely terminating translation of these proteins, resulting in non-functioning receptors that subsequently cause color blindness and nephrogenic diabetes insipidus, respectively. Robinson and colleagues (P. R. Robinson et al., Neuron 9, 1992, pp. 719-725) characterized the first mutation in a human G-protein coupled receptor that resulted in constitutive activation of the receptor and caused human disease. They found that when the amino acid Lys296 was mutated to Glu in the visual pigment rhodopsin, it was able to activate the G-protein transducin in the absence of light (its natural “agonist”). This particular mutation causes a particularly severe phenotype of retinitis pigmentosa (T. J. Keen et al., Genomics 11, 1991, pp. 199-205).
The number of constitutively active receptors that cause human disease is expanding. Multiple endocrinological and oncological disorders are caused by mutations that give rise to constitutively active receptors. These mutations have been shown to occur as a result of both spontaneous somatic events and as inherited germ line defects. A single point mutation in the luteinizing hormone receptor (Asp578-Gly), which causes male-linked precocious puberty, has been shown to be familial in caucasian populations (A. Shenker et al., Nature 365, 1993, pp. 652-654) and sporadic in Japanese populations (K. Yano et al., J. Clin. Endocrin. Metab. 79, 1994, pp. 1818-1823). Two different point mutations in the parathyroid hormone receptor confer constitutive activity and cause Jansen's metaphyseal chondroplasia (E. Schipani et al., New Eng. J. Med. 335, 1996, pp. 708-714; E. Schipani et al., Science 268, 1995, pp. 98-100). Furthermore, two activating mutations were found in the thyrotropin receptor, both of which were found to cause many sporadic thyroid adenomas (J. Parma et al., Nature 365, 1993, pp. 649-651). Taken together, these data attest to the widespread biological significance of constitutively active receptors and their role in human disease. It is, therefore, highly likely that constitutively active G-protein coupled receptors exist in the human nervous system and mutations in these neurotransmitter receptors, including the 5-HT2A receptor, may cause human neuropsychiatric. disease.
Constitutive activity has been described for a growing number of G-protein coupled neurotransmitter receptors. The dopamine D2 receptor has been reported to be constitutively active, and some antipsychotic compounds have been described as inverse agonists, although many of these compounds appear to be classical antagonists (Nilsson, C. L., et al., Neuropsychopharmacology 15, 1996, pp. 53-61; Hall, D. A. and Strange, P. G., Brit. J Pharm., 121, 1997, pp. 731-736) Similarly, of the thirteen known serotonin receptor subtypes, only three have been shown to possess constitutive activity, the 5-HT1A, 5-HT1D and 5-HT2C receptors. For example, E. L. Barker et al. (J. Biol. Chem. 269, 1994, pp. 11687-11690) describe an in vitro assay in which the wild-type 5-HT2C receptor displays constitutive activity. They further report that certain classically defined antagonists of the receptor, actually act as inverse agonists.
The creation of an activated 5-HT2A receptor by mutagenesis was recently described (Egan, C., T., et., al., J. Pharm Exp. Ther. 286(1), 1998, pp. 85-90). Aitering amino acid 322 from the wild type cysteine to lysine, glutamate, or arginine created activated 5-HT2A receptor mutants. This amino acid was chosen because it is analogous to the activating mutation produced in the α1b receptor (Kjelsberg, M. A., et al., J. Biol. Chem. 267(3), 1992, pp. 1430-1433). The activated 5-HT2A receptor displayed measurable constitutive activity, and six antipsychotics were shown to be inverse agonists (Egan, C. T., ibid.; and Egan, C. T., et al., Annals N.Y. Acad. Sci., 1999, pp. 136-139). These authors were unable to measure the constitutive activity of the wild type receptor in their assay, and an insufficient number of clinically relevant compounds comprising the various chemical classes of antipsychotics were tested. This precluded the authors from recognizing the significance of 5-HT2A receptor inverse agonism and efficacy as an antipsychotic.